Uhh.. No time to sleep cos if ya sleep ya don't eat... Gotta hold heat just to make ends meet.. - T3

I got to thinking about flies again, partially because Ron Davis came and gave a talk at our department and partially because I saw a couple papers in Nature showing sleep regulation is mediated by protein kinase A (PKA) signaling in the mushroom bodies. You may be surprised to find out that flies sleep, but they at least do something like sleeping where they sit very still and it's harder to arouse them. Then once I started digging in I found yet another paper in Current Biology that came out within two days of the Nature papers implicating serotonin 5-HT1A receptors in the mushroom bodies. 5-HT1A receptors and PKA are linked through the cyclic AMP (cAMP) signaling pathway. PKA is also referred to as cAMP-dependent protein kinase.

Where else have I seen all these cats before? Ah yes. Here they are inititating memory formation in Kandel's Nobel lecture about the mechanisms of long-term facilitation in another invertebrate (the sea slug).

In fact, the PKA signaling pathway in the mushroom bodies is known to be of major importance for memory formation in fruit flies as well. Two of the earliest discovered fly memory mutants, rutabaga and dunce, ended up lacking PKA pathway proteins, and at least the rutabaga memory phenotype can be rescued by inducing the missing protein just in adult mushroom bodies.

I keep talking about mushroom bodies. Mushroom bodies are a bilateral tripartite structures found in the insect brain. Most of the mushroom bodies are made up of the axons of Kenyon cells. The Kenyon cells are all collected together in a big round hunk of tissue positioned to the dorsal-posterior. A tract called the peduncle shoots down and forward and a little lateral and then abruptly splits into lobes. For simplicity's sake I will pretend that there are only alpha, beta, and gamma lobes. The alpha lobe shoots straight up from the branching point producing a vertical lobe. The horizontal offshoot from the branching point goes medial and consists of beta and gamma lobes which skilled fly neuroanatomists can tell apart using arcane fly researcher magic. Actually I think the gamma lobe is just like in front of the beta lobe or something.

Mushroom bodies are known to be important for memory (especially long-term memory) because of things like the aforementioned rutabaga rescue experiments and because you can screw up memory by cooling them or ablating them with hydroxyurea. Also, they are kind of in a good spot to encode the type of memory most people study in fruit flies (aversive olfactory conditioning). They are two or three synapses from the olfactory receptors and they receive neuromodulatory input from transmitters proposed to be involved in shock representation.

All of these papers use the Gal4-UAS system, so I ought to lay it out. If fly researchers want to affect a gene's expression in a cell-type specific fashion they can just pull a fly Gal4 line off the shelf. Gal4 is a transcription factor that has been inserted in multiple spots in the fly genome producing flies that make Gal4, say, just in certain Kenyon cells or in the antennal lobe, or in the protocerebrum, or wherever. This happens because the Gal4 sequence is being driven by some other gene's regulatory region (promoter). A lot of times, they don't even know which gene's promoter is being co-opted, but hey it works. Gal4 has the property of activating transcription of genes that are controlled by an Upstream Activating Sequence (UAS). So you have to make a fly that has whatever gene you're interested in controlled by a UAS. Then you grab a Gal4 strain that has expression in the cell-type you're interested in and cross'em. Gal4 + UAS = your gene expressed solely in the tissue of interest.

Pitman et al. started out by inserting a temperature-sensitive inhibitor of neurotransmitter release (shibire) after a UAS and trying it out in different Gal4 lines to find sleep centers. They found a few lines with reduced sleep (short-sleep lines, SSLs), and then looked to see where Gal4 was being expressed by using the same Gal4 lines but crossing them with UAS-GFP strains instead of the neural silencer. They mostly found expression in the mushroom bodies, so they tried out a couple more known mushroom body lines (c309 and 30Y), and these showed short-sleep phenotypes too. The overall idea here is that mushroom body activity is sleep-promoting. Interestingly, a c309-PKA line also reduced sleep. So turning off neurotransmitter release seems to have the same effect as turning up PKA activity.

Also, sleep-reducing changes were associated with a reduction in lifespan. I will use this finding to cynically sensationalize any further findings regarding sleep regulation by making them seem a matter of life or death. Since the sleep regulation machinery overlaps so much with synaptic plasticity and memory pathways, I will use this as an excuse to say things like "Learning can kill you!" or "Learning can save your life!" depending on which way I interpret the rest of this data.

Joiner et al. did a similar set of experiments, but instead of turning off neurotransmitter release in certain Gal4 lines, they expressed a constitutively active version of PKA. They found sleep phenotypes in two lines: 201Y and c309 again. It's rare to see an experimental result independently replicated period, let alone on the next page of the same journal, but here we have it. c309-PKA flies showed reduced sleep in Joiner et al.'s hands too. 201Y-PKA flies showed an increased sleep phenotype. The authors of this paper were a little more subtle with their histology and determined that c309 and 201Y did indeed both have mushroom body expression patterns, but they have a complementary pattern within the alpha and beta lobes. 201Y is found in gamma lobes and the core region of the alpha and beta lobes while c309 is found in the gamma lobes and not in the core region of the alpha and beta lobes.

These folks also have a mushroom body Gal4 line with similar expression pattern to c309 that can be turned on in response to a drug (RU486), so they could isolate the adult contribution rather than the developmental contribution of PKA activity. This manipulation was also associated with reduced sleep. Tissue and time-specifically expressing potassium channels to inhibit mushroom body activity or sodium channels to activate led to increased and decreased sleep respectively. But didn't we just learn that mushroom body activity is sleep-promoting? In fact, this paper also contains the finding that mushroom bodies leads to less sleep, but they note that other manipulations can produce much larger changes. Mushroom bodies probably contain populations that do both and when you ablate them you are canceling out a little bit of both.

I got interested in this idea and started looking for indicators of what might be different between the two populations. The major distinction seems to be that PKA activity (and thus more activity?) in the core of the alpha and beta lobes is sleep promoting while PKA outside of the core is sleep reducing. I found one paper that shows a GABA receptor subunit is primarily found in the core of alpha and beta lobes. GABA is normally an inhibitory input, so having GABA receptors throws a minus sign in the equation, but these are receptors not neurotransmitter, so this only explains a differential response to external stimuli between core and non-core. Another paper shows that calcium/calmodulin-dependent kinase II (CaMKII, an important synaptic plasticity molecule) is mostly non-core, as are two important PKA signaling molecules, Leonardo and Rutabaga. This indicates to me that it is a little unnatural to be expressing active PKA in the core (and thus promoting sleep). This leads me to believe that endogenous PKA activity is probably sleep reducing.

Finally, there's this Yuan et al. paper from the same lab as the Joiner et al. paper. In this paper the focus is on serotonin receptor subtypes. The authors produced disruptions in each of the drosophila 5HT receptor subtypes and looked for sleep phenotypes. Only the 5HT-1A receptor disruption had an effect on sleep, producing a short sleep phenotype. Reinstating 5HT-1A using a pan-neuronal Gal4 driver and UAS-5HT-1AR, rescued sleep. Also, expressing the receptor using the Gal4 MB-Switch driver (the one that can be turned on with RU486) just in adult mushroom bodies rescued sleep. This pretty much ruins my core versus non-core theorizing because the aforementioned "pan-neuronal" driver has more intense expression in the core while the MB-Switch driver is preferentially non-core, but the outcome is the same for expressing 5HT-1A receptors in both.

When I read all of these at once I had hoped to come up with some grand overarching theory of memory, sleep, and cyclic AMP signaling, but, alas, the data isn't that neatly sewn up. It seems like serotonin and PKA are having opposite effects. The story would fit more neatly into my scheme if PKA activity in the mushroom bodies was sleep promoting, and it is in certain neuronal populations. If I combine this with Pitman et al.'s general finding that mushroom body activity seems to be sleep-promoting and ignore the inconvenient data points from Joiner et al., then I could say something like "learning induces signal cascades that produce greater mushroom body activity and promote sleep." Then I could tie it in with theories about the role of network activity (like sharp-wave ripples) in the mammalian brain during sleep. Turns out it isn't that simple. Any ideas? How do you like reading a blog post without the answer at the end?